Keyword research is how you decide what to write before you write it: the work of finding the actual phrases people type into search engines, then scoring them by intent, demand, and difficulty so you only build pages worth building. Done right, it stops you from publishing things nobody searches for and points you at the gaps your competitors left open. We treat it as the first move in any SEO program, and we keep it running long after the first spreadsheet is done.
Keyword Research
Keyword research is the process of discovering, analyzing, and prioritizing the search terms your audience uses, so you can target the queries with the best balance of intent, demand, and winnability.
Why it matters
A keyword is a proxy for a person and a problem. When you match a query to a page, you’re really matching a searcher’s intent to something useful you can offer. Skip the research and you’re guessing: building pages on vibes, optimizing for terms with no demand, or chasing head terms your domain can’t touch yet.
Good keyword research does three jobs at once:
- Reveals intent. Every query carries a job-to-be-done. “best CRM for agencies” wants a comparison; “what is a CRM” wants a definition; “hubspot login” wants a door. Map the query to the right page type and you stop bouncing traffic you fought to win.
- Sizes the opportunity. Volume, trend, and commercial intent tell you whether a topic is worth the editorial cost. A 50-search-a-month term that converts at 8% often beats a 5,000-search term that converts at nothing.
- Exposes the gaps. Competitor keyword analysis shows where rivals rank and where they don’t; the unclaimed queries are your fastest path to traffic.
What comes out the other end is a prioritized map of which page targets which cluster, and it feeds your site architecture, internal linking, and editorial calendar.
The workflow we actually run
One tool and a vibe is how most keyword lists get made. Ours is a repeatable pipeline.
1. Seed the topic
Start with what you sell and the problems you solve, then expand. Pull seed terms from your own analytics and Search Console (queries you already get impressions for), site search logs, sales-call transcripts, and the language customers use in reviews and support tickets. Internal data is the most underused seed source there is: it’s intent you’ve already paid to collect.
2. Expand into the full demand space
Feed seeds into a keyword tool to harvest matching terms, questions, and related phrases. Mine the SERP itself: People Also Ask, autocomplete, “related searches,” and the SERP features Google chooses to show. For deep product or category work, layer in PPC keyword research: Google Ads Keyword Planner bid ranges and your paid search terms report expose commercial terms that organic tools underweight.
3. Classify by intent
Tag every term against the four classic intent buckets. This is the step most people skip, and it’s the one that decides whether a page ranks and converts.
| Intent | Searcher wants | Example query | Page type that wins |
|---|---|---|---|
| Informational | To learn | ”how does keyword research work” | Guide, glossary, blog |
| Commercial | To compare before buying | ”best keyword research tools” | Comparison, listicle, review |
| Transactional | To act / buy now | ”ahrefs pricing” | Product, pricing, signup |
| Navigational | A specific brand/page | ”semrush login” | Brand homepage, login |
4. Score and prioritize
For each term, weigh volume against difficulty against business value. Don’t optimize for volume alone: a new site chasing high-difficulty head terms burns months for nothing. Favor low-competition keywords and long-tail phrases where you can win this quarter, then climb toward the harder terms as authority accrues.
5. Cluster the list
Group related terms into topics. One pillar page can target a whole cluster of intent-adjacent queries; ten thin pages targeting near-duplicate phrases just cannibalize each other. This is where topic clusters and semantic SEO turn a flat keyword list into a content architecture.
6. Map keywords to pages
Assign each cluster a primary keyword and a target URL. Decide which existing page gets optimized and which gaps need net-new pages. This mapping is the bridge between research and production, and the spine of any SEO campaign we launch.
No keyword survives contact with the SERP unexamined. Before committing to a target, read the page-one results: if they’re all 3,000-word ultimate guides and you planned a 400-word post, the SERP is telling you the price of entry.
Keyword research in the AI Overviews era
Search stopped being ten blue links a while ago. AI Overviews, AI Mode, and answer engines now absorb a large share of informational clicks, which changes the math on which terms are worth chasing for traffic versus visibility. Pure “what is X” definitional queries increasingly get answered on the SERP without a click, and comparison queries carry AI Overviews too: in our check of 30 SERPs, “best mattress” and “crm software” had one, while transactional, tool and local queries mostly did not. On most of those SERPs, a citation inside the AI answer is now a target of its own, next to the blue-link slot.
Practically, that means: weight your prioritization toward intent and conversion fit over raw volume, target the question-shaped long-tail that AI summaries pull from, and build the topical depth and entity coverage that makes you the source an answer engine quotes. Your own data matters more here too. Third-party volumes are modeled estimates, and Search Console withholds rare queries to protect searchers’ privacy, so the queries you already earn impressions for are the most reliable signal of what’s actually working.
Common mistakes to avoid
- Relying solely on search volume. High volume doesn’t guarantee relevance or conversions, and it counts searches, which is a bigger number than clicks: in our 30-keyword study, volume overstated organic clicks about twofold, and most on commercial terms. Combine volume with intent, CTR, and conversion data.
- Ignoring search intent. Targeting terms that don’t match what the searcher wants drives bounces. Map each keyword to informational, commercial, transactional, or navigational intent.
- Skipping long-tail keywords. Chasing only head terms misses qualified traffic. Add long-tail and question phrases to capture specific queries with lower competition.
- Trusting a single tool. One tool introduces bias. Cross-check multiple tools, the live SERP, and your own analytics.
- Ignoring SERP features and competition. Overlooking featured snippets, People Also Ask, and top pages leads to bad targets. Audit the SERP before committing.
- Chasing terms beyond your authority. High-difficulty terms waste resources for a new site. Balance difficulty against your domain strength and win the winnable first.
- Keyword-stuffing. Cramming terms unnaturally hurts UX and rankings, and inflates keyword density past the point of usefulness. Use terms naturally and prioritize topical coverage.
- Treating keywords in isolation. Fragmented targeting splinters authority. Cluster related terms into pillar and supporting pages.
- Ignoring internal data. Site search, analytics, and Search Console hold proven queries. Mine them before you mine any third-party tool.
- Forgetting seasonality. Static lists miss demand swings. Track trends and plan content around peaks.
- Over-indexing on exact match. Modern ranking favors topical relevance. Optimize for the semantic variations and related questions around a phrase.
- Pruning nothing. Outdated targets accumulate. Review performance and cut or refresh low-value keywords on a schedule.
Putting it to work
Keyword research is only useful when it ends in shipped, mapped pages. Once you have a clustered list, validate the targets against the live SERP, map them to URLs, and feed the gaps into an SEO content audit of what you already have. At scale the stakes rise: on a programmatic SEO build, research is the input that decides whether a thousand generated pages earn traffic or just bloat the index.
When keyword research becomes a business problem
You have a 2,000-row export sorted by volume and no idea which 20 pages to build. The blog ships every week, impressions climb, and demo requests stay where they were. Two of your own pages fight over the same query, and the terms you picked are owned by sites with a decade of links.
If you sell one product into a market with a few dozen real queries, the six steps above, your own Search Console and a free tool tier will cover it.
When the list runs into the thousands, a B2B SEO engagement runs the research against pipeline: intent labels, a read of each SERP, clusters mapped to URLs, and the terms that will never produce a lead cut early. RMIQ is the example: high-intent buyer queries in a retail-media category that barely existed in search, and about 285,000 Google impressions a month at peak.
Frequently Asked Questions
What is keyword research in SEO?
Keyword research is the process of finding and analyzing the search terms your audience uses, then prioritizing them by intent, search volume, and difficulty. It tells you which queries are worth building pages for, so your content targets demand you have actually measured, and it’s the foundation of any SEO or content strategy.
How do I find keywords for my website?
Start with seed terms from what you sell and your own Search Console and site-search data. Expand them in a keyword tool to harvest related terms and questions, then mine the live SERP: People Also Ask, autocomplete, related searches. Classify each term by intent, score it by volume and difficulty, and cluster the survivors into page targets.
What is the difference between short-tail and long-tail keywords?
Short-tail (head) keywords are broad, high-volume terms like “shoes”, competitive and rarely specific about intent. Long-tail keywords are longer, lower-volume phrases like “waterproof trail running shoes for wide feet”, easier to rank for and far clearer about what the searcher wants, which usually means they convert better.
Is keyword research still relevant with AI Overviews?
Yes, and the emphasis shifts. AI Overviews answer many definitional queries without a click, and they now appear on comparison queries too, so prioritize terms with buying intent and the question-shaped long tail that AI summaries pull from. The goal expands from ranking a blue link to becoming the source answer engines cite, which still starts with knowing exactly what people search for.
What is PPC keyword research?
PPC keyword research is the same discovery work done for paid search: finding the queries worth bidding on, grouping them into ad groups, and adding negative keywords for the ones you never want to pay for. Its data is useful for SEO too. Bid ranges in Google Ads Keyword Planner show which terms advertisers value most, and your search terms report lists the exact queries that already produced paid clicks and conversions.
How often should I redo keyword research?
Treat it as continuous. Review performance and refresh targets quarterly, and revisit sooner when you launch products, enter new markets, or hit seasonal peaks. Your own Search Console data should feed back constantly: new impression-earning queries are free signals about where the next opportunity is.